What NYC campaigns can learn from ranked choice elections in other cities

March 10, 2021
Since 2004, ranked choice voting (RCV) has taken off in dozens of jurisdictions across the country. Multiple cities in the Bay Area, both Twin Cities, and the entire state of Maine have all rolled out their own versions of ranked choice voting. But this June, the population of voters eligible to cast their ballots via […]

Since 2004, ranked choice voting (RCV) has taken off in dozens of jurisdictions across the country. Multiple cities in the Bay Area, both Twin Cities, and the entire state of Maine have all rolled out their own versions of ranked choice voting.

But this June, the population of voters eligible to cast their ballots via RCV will nearly double as New York City enters the fold.

Across the jurisdictions that have rolled out RCV in the past, the warm up period with ranked choice voting has been at least a little bumpy. Since we’ve had some experience supporting campaigns in these elections, we thought it might be helpful to 1) respond to some common questions we hear from campaigns and 2) reflect on some of the targeting lessons we’ve learned along the way.

Who votes in ranked choice elections?

This new system is not intuitive for all voters. Does that change how many people might vote?

The answer thus far has been not really. There’s some evidence that RCV results in higher turnout in primary elections, but in the end, the same drivers of turnout that have historically been important remain the leading factors. Those factors include how competitive the elections are, how much money candidates spend, and how much the race is covered in the media.

However, there are two interesting points:

In our models, we see a pretty clear relationship between the number of major campaigns in a race and the size of the electorate. Since this system — alongside other factors like term limits and public financing — has contributed to such a large number of candidates on the New York City ballot in June, turnout could be pretty high.

In 2014, a study looking at RCV in Minneapolis found while the total share of people voting is not that different under RCV, the composition of the electorate is. The authors found that RCV increases turnout among white voters and voters with high incomes while lowering turnout among voters of color and voters with low incomes. However, this effect appears to fade over time as voters become more familiar with the system.

Do voters fill out their ballots correctly?

This system asks voters to rank their preferred candidates for each office. In most jurisdictions, you get three ranks. But NYC is going big with five ranks for each office. That means voters will indicate their five most preferred candidates for each office in order.

In San Francisco, which has three ranks, most voters complete their ballots. But more than a third of voters do not — even in competitive elections. From 2004 to today, about 61% of voters in San Francisco have completed their ballots in races with more than three candidates. Ballot completion tends to be higher in more high-profile, citywide races and lower in races further down the ballot.

These trends have been pretty consistent across the country. Ballot completion improves over time as people get more familiar with the new system, but we’ve seen that the first few cycles benefit from big investments in voter education.

Should I team up with other campaigns?

We worked in the 2016 Oakland election, a few cycles after RCV was first adopted there. Digging through historic data to build our models, we noticed something interesting and surprising: in the first years of ranked choice voting, there was no significant correlation between support for specific candidates in most races. It’s the complete opposite of what we expected!

But as the system became more familiar to voters and campaigns, those correlations began to emerge. And in more recent RCV races, we’ve begun to see pretty strong correlations in support for ideologically similar candidates.

We believe the main factors behind the emergence of this correlation have been 1) spending by outside groups supporting or opposing blocks of candidates (rather than individuals), 2) cooperative campaigning among groups of candidates, and 3) increased media coverage of multiple campaigns as a block.

Mark Leno and Jane Kim campaigned as a slate in the 2018 San Francisco Mayor’s race. Kim’s ads advocated ranking her first and Leno second. Leno’s ads advocated the inverse.

Most voters are not as engaged in politics as those of us involved in campaigns. As a result, they’re going to have a hard time keeping track of all the candidates they’re supposed to be ranking and will need help making sense of which candidates belong together.

Will NYC react to RCV differently than other jurisdictions?

There’s never been an American RCV experiment like the one we’re about to see in New York.

There’s an unprecedented number of candidates, an unprecedented amount of money being raised and spent, and an unprecedented level of media coverage for an RCV election. So who knows what the heck will happen!

Look at all these damn candidates!!!

All the lessons from other cities may go out the window, but they’re the best we’ve got at this point.

Targeting takeaways

From a targeting perspective, we see a few key takeaways:

Focus new programs on low-propensity voters

Normally, campaigns think about turnout as an exercise in mobilizing their likely supporters who have mid-range turnout probabilities. That’s because high propensity voters don’t need your help and low propensity voters are often very hard to mobilize.

Given the turnout polarization we’ve seen in other RCV districts — where lower propensity voters become even less likely to vote — it’s worth asking whether those assumptions regarding how to target your turnout program should be challenged.

We think you can’t just abandon these lower propensity voters. Especially because their turnout is is likely lower in these elections because the new rules are confusing for folks not paying close attention. That’s a fixable problem!

So we recommend doing the work of reaching even lower propensity voters than you normally would in a turnout program, with messaging focused on voter education. For example, you might send voters sample ballots in the mail, send patient training-oriented volunteers to knock on their doors, and serve them digital ads linking to a website where they could click around a sample ballot and get to know the system better.

Form coalitions as quickly as possible

We’ve seen that without coalitions, the benefits campaigns might get out of the ranking system fall apart a bit. And that’s because coalitions are critical to educating voters on who to rank together — regardless of order.

It might feel like a weird vibe to reach out to other campaigns like you’re teaming up in a game of RISK, but this is a different type of election and the sooner campaigns adapt, the sooner they benefit.

Help your allies’ voters understand the ballot

If you have reason to believe that a certain opponent’s supporters are also inclined to support you (perhaps due to coalition campaigning), you should engage those voters in the same kinds of voter education efforts used in your turnout program.

But here, you might focus your targeting on those most likely to vote and focus your messaging on the value of successfully completing your ballot. Even reliable voters leave their ballots incomplete in pretty large numbers.

Focus on positive persuasion (with exceptions!)

Given the randomness we’ve seen in secondary votes in early RCV elections in other cities, you probably don’t want to alienate other candidates’ supporters — even if you think that candidate is bad.

Making a positive case for your own campaign will increase your odds of getting ranked on more ballots, which should be your main goal in persuasion. Even if there’s another candidate clearly beating you in first round voting who you think you need to take down, you can take them down instead by getting more secondary votes from the primary voters of candidates likely to be eliminated.

But there are exceptions. If you are in a very crowded race and you believe your campaign is near the bottom of the pack, a more aggressive approach focused on why your campaign is a better pick than the leaders in your race probably makes sense. Also, if you believe a candidate in your race is likely to clear 50% of the vote in the first round, persuasion should be focused on pulling support away from that candidate.

What we’re doing to help

Given the importance of building coalitions and persuading audiences likely to rank you as a secondary option, we’ve introduced a new targeting filter just for ranked choice elections.

In RCV elections, you can now use Deck to target the likely supporters (or non-supporters) of other candidates in your race.

In addition to creating lists using turnout scores and your own campaign’s support scores, you can now also build target audiences using your opponents’ support scores.

For example, if you’re Maya Wiley, you could create a persuasion audience focused on likely first rank Morales voters. And you could create a different persuasion audience focused on first rank McGuire or Yang voters.

We’re hoping to build this out a bit more in the near future to recommend which candidates’ supporters are most aligned with yours as possible coalition partners.

We’re so excited to be supporting campaigns in the NYC primary. If you’re interested in chatting more about what we’ve discussed in the blog post or learning more about Deck, please sign up for a live demo!

More Like This